Abstract
Abstract Predicting gene expression from cis-regulatory DNA sequences at the promoter and terminator regions is a central challenge in plant genomics. This capability is also a prerequisite for assessing the effects of regulatory mutations on gene expression. Here, we developed deep learning sequence-to-expression (S2E) models that leverage context-aware sequence embeddings from the PlantCaduceus genomic language model instead of one-hot encoding of sequences, to predict gene expression across 17 plant species. To further improve predictions, we integrated chromatin accessibility data as auxiliary regulatory features. First, we evaluated our models to predict gene expression on unseen gene families via cross-validation, demonstrating our model’s prediction accuracy across all species outperforms PhytoExpr, the current state-of-the-art (SOTA) S2E model in plants (Pearson R=0.82 vs. R=0.74). We then validated variant effect predictions using an experimental dataset across 796 Brachypodium mutant lines, specifically designed to test predictions at single-base resolution. Our models outperformed SOTA S2E models in predicting between-gene expression differences (regression coefficient β=0.78 vs. β=0.57). Remarkably, they also accurately predicted the effects of single-nucleotide mutations on within-gene expression, while SOTA S2E models showed only weak associations (regression coefficient β=0.38 vs. β=0.08). Our results demonstrated the value of context-aware DNA sequence embeddings for predicting regulatory variant effects in plants. They also reveal a persistent accuracy gap in S2E models when moving from between-gene to allelic variation, a challenge that needs to be addressed in future S2E studies.
Abstract from the paper’s OpenAlex record; open copy licensed cc-by-nc. Read at the source ↗
The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.
Claims 4
Claim · model-written summary
The S2E model using PlantCaduceus embeddings predicted gene expression across plant species with R=0.82, outperforming PhytoExpr (R=0.74).
Verbatim
e1“demonstrating our model’s prediction accuracy across all species outperforms PhytoExpr, the current state-of-the-art (SOTA) S2E model in plants (Pearson R=0.82 vs. R=0.74)”
Claim · model-written summary
The S2E models outperformed SOTA models in predicting between-gene expression differences in Brachypodium mutant lines (beta 0.78 vs 0.57).
- Taxon
- Brachypodium
Brachypodiumt0 - Outcome
- predicting between-gene expression differencese2
- Direction
- outperformede2
- Value
- β=0.78e2
- Qualifier
- β=0.57e2
- Comparator
- SOTA S2E modelse2
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Brachypodium mutant lines”
e2“Our models outperformed SOTA S2E models in predicting between-gene expression differences (regression coefficient β=0.78 vs. β=0.57)”
Claim · model-written summary
The S2E models predicted effects of single-nucleotide mutations on within-gene expression in Brachypodium (beta 0.38 vs 0.08 for SOTA models).
- Taxon
- BrachypodiumBrachypodiumt0
- Outcome
- predicted the effects of single-nucleotide mutations on within-gene expressione3
- Value
- β=0.38e3
- Qualifier
- β=0.08e3
- Comparator
- SOTA S2E modelse3
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Brachypodium mutant lines”
e3“they also accurately predicted the effects of single-nucleotide mutations on within-gene expression, while SOTA S2E models showed only weak associations (regression coefficient β=0.38 vs. β=0.08)”
Claim · model-written summary
S2E models show a persistent accuracy gap when moving from between-gene to allelic variation.
- Taxon
- BrachypodiumBrachypodiumt0
- Outcome
- persistent accuracy gape4
- Comparator
- between-genee4
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“Brachypodium mutant lines”
e4“They also reveal a persistent accuracy gap in S2E models when moving from between-gene to allelic variation”